The LOGINIT, LOGINTR, LOGRM, and LOGPMU buffers are circular buffers
that have printf-like logs from GSP-RM and PMU encoded in them.
LOGINIT, LOGINTR, and LOGRM are allocated by Nouveau and their DMA
addresses are passed to GSP-RM during initialization. The buffers are
required for GSP-RM to initialize properly.
LOGPMU is also allocated by Nouveau, but its contents are updated
when Nouveau receives an NV_VGPU_MSG_EVENT_UCODE_LIBOS_PRINT RPC from
GSP-RM. Nouveau then copies the RPC to the buffer.
The messages are encoded as an array of variable-length structures that
contain the parameters to an NV_PRINTF call. The format string and
parameter count are stored in a special ELF image that contains only
logging strings. This image is not currently shipped with the Nvidia
driver.
There are two methods to extract the logs.
OpenRM tries to load the logging ELF, and if present, parses the log
buffers in real time and outputs the strings to the kernel console.
Alternatively, and this is the method used by this patch, the buffers
can be exposed to user space, and a user-space tool (along with the
logging ELF image) can parse the buffer and dump the logs.
This method has the advantage that it allows the buffers to be parsed
even when the logging ELF file is not available to the user. However,
it has the disadvantage the debugfs entries need to remain until the
driver is unloaded.
The buffers are exposed via debugfs. If GSP-RM fails to initialize, then
Nouveau immediately shuts down the GSP interface. This would normally
also deallocate the logging buffers, thereby preventing the user from
capturing the debug logs.
To avoid this, introduce the keep-gsp-logging command line parameter. If
specified, and if at least one logging buffer has content, then Nouveau
will migrate these buffers into new debugfs entries that are retained
until the driver unloads.
An end-user can capture the logs using the following commands:
cp /sys/kernel/debug/nouveau/<path>/loginit loginit
cp /sys/kernel/debug/nouveau/<path>/logrm logrm
cp /sys/kernel/debug/nouveau/<path>/logintr logintr
cp /sys/kernel/debug/nouveau/<path>/logpmu logpmu
where (for a PCI device) <path> is the PCI ID of the GPU (e.g.
0000:65:00.0).
Since LOGPMU is not needed for normal GSP-RM operation, it is only
created if debugfs is available. Otherwise, the
NV_VGPU_MSG_EVENT_UCODE_LIBOS_PRINT RPCs are ignored.
A simple way to test the buffer migration feature is to have
nvkm_gsp_init() return an error code.
Tested-by: Ben Skeggs <bskeggs@nvidia.com>
Signed-off-by: Timur Tabi <ttabi@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20241030202952.694055-2-ttabi@nvidia.com
Typically DRM drivers use the DRM debugfs root entry. However, since
Nouveau is heading towards a split into a core and a DRM driver, create
a module specific debugfs root directory.
Subsequent patches make use of this new debugfs root in order to store
GSP-RM log bufferes (optionally beyond a device driver binding).
Acked-by: Timur Tabi <ttabi@nvidia.com>
Signed-off-by: Danilo Krummrich <dakr@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20241125142639.9126-1-dakr@kernel.org
Remove the connector-status initialization from several of vmwgfx's
functions. It is not required by the driver or DRM helpers.
DRM initializes the connector to unknown status in __drm_connector_init()
and reads the physical status when DRM clients or the user needs the
information, or if it detects/polls a change to the status.
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Reviewed-by: Zack Rusin <zack.rusin@broadcom.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241105132045.471856-1-tzimmermann@suse.de
Pull drm fixes from Dave Airlie:
"Merge window fixes, mostly amdgpu and xe, with a few other minor ones,
all looks fairly normal,
i915:
- hdcp: Fix when the first read and write are retried
xe:
- Wake up waiters after wait condition set to true
- Mark the preempt fence workqueue as reclaim
- Update xe2 graphics name string
- Fix a couple of guc submit races
- Fix pat index usage in migrate
- Ensure non-cached migrate pagetable bo mappings
- Take a PM ref in the delayed snapshot capture worker
amdgpu:
- SMU 13.0.6 fixes
- XGMI fixes
- SMU 13.0.7 fixes
- Misc code cleanups
- Plane refcount fixes
- DCN 4.0.1 fixes
- DC power fixes
- DTO fixes
- NBIO 7.11 fixes
- SMU 14.0.x fixes
- Reset fixes
- Enable DC on LoongArch
- Sysfs hotplug warning fix
- Misc small fixes
- VCN 4.0.3 fix
- Slab usage fix
- Jpeg delayed work fix
amdkfd:
- wptr handling fixes
radeon:
- Use ttm_bo_move_null()
- Constify struct pci_device_id
- Fix spurious hotplug
- HPD fix
rockchip
- fix 32-bit build"
* tag 'drm-next-2024-11-29' of https://gitlab.freedesktop.org/drm/kernel: (48 commits)
drm/xe: Take PM ref in delayed snapshot capture worker
drm/xe/migrate: use XE_BO_FLAG_PAGETABLE
drm/xe/migrate: fix pat index usage
drm/xe/guc_submit: fix race around suspend_pending
drm/xe/guc_submit: fix race around pending_disable
drm/xe: Update xe2_graphics name string
drm/rockchip: avoid 64-bit division
Revert "drm/radeon: Delay Connector detecting when HPD singals is unstable"
drm/amdgpu/jpeg: cancel the jpeg worker
drm/amdgpu: fix usage slab after free
drm/amdgpu/vcn: reset fw_shared when VCPU buffers corrupted on vcn v4.0.3
drm/amdgpu: Fix sysfs warning when hotplugging
drm/amdgpu: Add sysfs interface for vcn reset mask
drm/amdgpu/gmc7: fix wait_for_idle callers
drm/amd/pm: Remove arcturus min power limit
drm/amd/pm: skip setting the power source on smu v14.0.2/3
drm/amd/pm: disable pcie speed switching on Intel platform for smu v14.0.2/3
drm/amdkfd: Use the correct wptr size
drm/xe: Mark preempt fence workqueue as reclaim
drm/xe/ufence: Wake up waiters after setting ufence->signalled
...
Pull driver core updates from Greg KH:
"Here is a small set of driver core changes for 6.13-rc1.
Nothing major for this merge cycle, except for the two simple merge
conflicts are here just to make life interesting.
Included in here are:
- sysfs core changes and preparations for more sysfs api cleanups
that can come through all driver trees after -rc1 is out
- fw_devlink fixes based on many reports and debugging sessions
- list_for_each_reverse() removal, no one was using it!
- last-minute seq_printf() format string bug found and fixed in many
drivers all at once.
- minor bugfixes and changes full details in the shortlog"
* tag 'driver-core-6.13-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core: (35 commits)
Fix a potential abuse of seq_printf() format string in drivers
cpu: Remove spurious NULL in attribute_group definition
s390/con3215: Remove spurious NULL in attribute_group definition
perf: arm-ni: Remove spurious NULL in attribute_group definition
driver core: Constify bin_attribute definitions
sysfs: attribute_group: allow registration of const bin_attribute
firmware_loader: Fix possible resource leak in fw_log_firmware_info()
drivers: core: fw_devlink: Fix excess parameter description in docstring
driver core: class: Correct WARN() message in APIs class_(for_each|find)_device()
cacheinfo: Use of_property_present() for non-boolean properties
cdx: Fix cdx_mmap_resource() after constifying attr in ->mmap()
drivers: core: fw_devlink: Make the error message a bit more useful
phy: tegra: xusb: Set fwnode for xusb port devices
drm: display: Set fwnode for aux bus devices
driver core: fw_devlink: Stop trying to optimize cycle detection logic
driver core: Constify attribute arguments of binary attributes
sysfs: bin_attribute: add const read/write callback variants
sysfs: implement all BIN_ATTR_* macros in terms of __BIN_ATTR()
sysfs: treewide: constify attribute callback of bin_attribute::llseek()
sysfs: treewide: constify attribute callback of bin_attribute::mmap()
...
Currently in some testcases we can trigger:
xe 0000:03:00.0: [drm] Assertion `exec_queue_destroyed(q)` failed!
....
WARNING: CPU: 18 PID: 2640 at drivers/gpu/drm/xe/xe_guc_submit.c:1826 xe_guc_sched_done_handler+0xa54/0xef0 [xe]
xe 0000:03:00.0: [drm] *ERROR* GT1: DEREGISTER_DONE: Unexpected engine state 0x00a1, guc_id=57
Looking at a snippet of corresponding ftrace for this GuC id we can see:
162.673311: xe_sched_msg_add: dev=0000:03:00.0, gt=1 guc_id=57, opcode=3
162.673317: xe_sched_msg_recv: dev=0000:03:00.0, gt=1 guc_id=57, opcode=3
162.673319: xe_exec_queue_scheduling_disable: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0x29, flags=0x0
162.674089: xe_exec_queue_kill: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0x29, flags=0x0
162.674108: xe_exec_queue_close: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0xa9, flags=0x0
162.674488: xe_exec_queue_scheduling_done: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0xa9, flags=0x0
162.678452: xe_exec_queue_deregister: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0xa1, flags=0x0
It looks like we try to suspend the queue (opcode=3), setting
suspend_pending and triggering a disable_scheduling. The user then
closes the queue. However the close will also forcefully signal the
suspend fence after killing the queue, later when the G2H response for
disable_scheduling comes back we have now cleared suspend_pending when
signalling the suspend fence, so the disable_scheduling now incorrectly
tries to also deregister the queue. This leads to warnings since the queue
has yet to even be marked for destruction. We also seem to trigger
errors later with trying to double unregister the same queue.
To fix this tweak the ordering when handling the response to ensure we
don't race with a disable_scheduling that didn't actually intend to
perform an unregister. The destruction path should now also correctly
wait for any pending_disable before marking as destroyed.
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/3371
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: <stable@vger.kernel.org> # v6.8+
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241122161914.321263-6-matthew.auld@intel.com
(cherry picked from commit f161809b36)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Currently in some testcases we can trigger:
[drm] *ERROR* GT0: SCHED_DONE: Unexpected engine state 0x02b1, guc_id=8, runnable_state=0
[drm] *ERROR* GT0: G2H action 0x1002 failed (-EPROTO) len 3 msg 02 10 00 90 08 00 00 00 00 00 00 00
Looking at a snippet of corresponding ftrace for this GuC id we can see:
498.852891: xe_sched_msg_add: dev=0000:03:00.0, gt=0 guc_id=8, opcode=3
498.854083: xe_sched_msg_recv: dev=0000:03:00.0, gt=0 guc_id=8, opcode=3
498.855389: xe_exec_queue_kill: dev=0000:03:00.0, 5:0x1, gt=0, width=1, guc_id=8, guc_state=0x3, flags=0x0
498.855436: xe_exec_queue_lr_cleanup: dev=0000:03:00.0, 5:0x1, gt=0, width=1, guc_id=8, guc_state=0x83, flags=0x0
498.856767: xe_exec_queue_close: dev=0000:03:00.0, 5:0x1, gt=0, width=1, guc_id=8, guc_state=0x83, flags=0x0
498.862889: xe_exec_queue_scheduling_disable: dev=0000:03:00.0, 5:0x1, gt=0, width=1, guc_id=8, guc_state=0xa9, flags=0x0
498.863032: xe_exec_queue_scheduling_disable: dev=0000:03:00.0, 5:0x1, gt=0, width=1, guc_id=8, guc_state=0x2b9, flags=0x0
498.875596: xe_exec_queue_scheduling_done: dev=0000:03:00.0, 5:0x1, gt=0, width=1, guc_id=8, guc_state=0x2b9, flags=0x0
498.875604: xe_exec_queue_deregister: dev=0000:03:00.0, 5:0x1, gt=0, width=1, guc_id=8, guc_state=0x2b1, flags=0x0
499.074483: xe_exec_queue_deregister_done: dev=0000:03:00.0, 5:0x1, gt=0, width=1, guc_id=8, guc_state=0x2b1, flags=0x0
This looks to be the two scheduling_disable racing with each other, one
from the suspend (opcode=3) and then again during lr cleanup. While
those two operations are serialized, the G2H portion is not, therefore
when marking the queue as pending_disabled and then firing off the first
request, we proceed do the same again, however the first disable
response only fires after this which then clears the pending_disabled.
At this point the second comes back and is processed, however the
pending_disabled is no longer set, hence triggering the warning.
To fix this wait for pending_disabled when doing the lr cleanup and
calling disable_scheduling_deregister. Also do the same for all other
disable_scheduling callers.
Fixes: dd08ebf6c3 ("drm/xe: Introduce a new DRM driver for Intel GPUs")
Closes: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/3515
Signed-off-by: Matthew Auld <matthew.auld@intel.com>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: <stable@vger.kernel.org> # v6.8+
Reviewed-by: Matthew Brost <mattheq.brost@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241122161914.321263-5-matthew.auld@intel.com
(cherry picked from commit ddb106d212)
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
It is permitted for a plane to be configured such that none
of it is on-screen via either negative dest rectangle X,Y
offset, or an offset that is greater than the crtc dimensions.
These planes were resized via drm_atomic_helper_check_plane_state
such that the source rectangle had a zero width or height, but
they still created a dlist entry even though they contributed
no pixels. In the case of vc6_plane_mode_set, that it could result
in negative values being written into registers, which caused
incorrect behaviour.
Drop planes that result in a source width or height of 0 pixels
or an on-screen size of 0 pixels to avoid the incorrect rendering.
Reviewed-by: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20241025-drm-vc4-2712-support-v2-28-35efa83c8fc0@raspberrypi.com
Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
